124218 - if kCSSRawKeywords gets out of sync with eCSSKeyword_COUNT nsCSSKeywords::AddRefTable might crash in DEBUG code

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, enhancement)

Description

[timeless]

steps, apply this very broken patch:
Index: nsCSSKeywords.cpp
===================================================================
RCS file: /cvsroot/mozilla/content/shared/src/nsCSSKeywords.cpp,v
retrieving revision 3.20
diff -u -0 -r3.20 nsCSSKeywords.cpp
--- nsCSSKeywords.cpp   2001/12/17 07:11:20     3.20
+++ nsCSSKeywords.cpp   2002/02/07 19:36:10
@@ -47 +46,0 @@
-#include "nsCSSKeywordList.h"

#7  0x28f0febe in nsCSSKeywords::AddRefTable () at /home/timeless/mozilla/content/shared/src/nsCSSKeywords.cpp:64
64              nsCAutoString temp1(kCSSRawKeywords[index]);
(gdb) l
59          if (gKeywordTable) {
60      #ifdef DEBUG
61          {
62            // let's verify the table...
63            for (PRInt32 index = 0; index < eCSSKeyword_COUNT; ++index) {
64              nsCAutoString temp1(kCSSRawKeywords[index]);
65              nsCAutoString temp2(kCSSRawKeywords[index]);
66              ToLowerCase(temp1);
67              NS_ASSERTION(temp1.Equals(temp2), "upper case char in table");
68              NS_ASSERTION(-1 == temp1.FindChar('_'), "underscore char in table");

(gdb) p index
$5 = 187
(gdb) p kCSSRawKeywords[187]
$14 = 0xa5d12e0 <Error reading address 0xa5d12e0: Bad address>

(gdb) p kCSSRawKeywords[1]
$6 = 0x0
(gdb) p kCSSRawKeywords[2]
$7 = 0x0
(gdb) p kCSSRawKeywords[3]
$8 = 0x290e3cf1 "-moz-appearance"
(gdb) p kCSSRawKeywords[4]
$9 = 0x290e3d01 "-moz-border-radius"
(gdb) p kCSSRawKeywords[5]
$10 = 0x290e3d14 "-moz-border-radius-topleft"
(gdb) p kCSSRawKeywords[6]
$11 = 0x290e3d2f "-moz-border-radius-topright"
(gdb) p kCSSRawKeywords[7]
$12 = 0x290e3d4b "-moz-border-radius-bottomleft"
(gdb) p kCSSRawKeywords[8]
$13 = 0x290e3d80 "-moz-border-radius-bottomright"
(gdb) p kCSSRawKeywords[186]
$15 = 0x0

feel free to mark invalid, otherwise if you're willing to accept a donated patch that would null check walking the array reassign to me.

Comment 1

[Boris Zbarsky [:bzbarsky]]

confirming to get this off the unco list...

Comment 2

[Christopher Hoess (gone)]

ok, reassigning

Comment 3

[timeless]

.

Comment 4

[timeless]

Attachment: patch

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 133841 [details] [diff] [review]
patch

r+sr=bzbarsky.	Looks reasonable.

Comment 6

[timeless]

checked in